PEEB / utils /load_model.py
Peijie's picture
load model in cpu first
c8636e8
raw
history blame
1.57 kB
try:
import spaces
gpu_decorator = spaces.GPU
except ImportError:
# Define a no-operation decorator as fallback
def gpu_decorator(func):
return func
import torch
from transformers import OwlViTProcessor, OwlViTForObjectDetection
from .model import OwlViTForClassification
def load_xclip(device: str = "cuda:0",
n_classes: int = 183,
use_teacher_logits: bool = False,
custom_box_head: bool = False,
model_path: str = 'data/models/peeb_pretrain.pt',
):
owlvit_det_processor = OwlViTProcessor.from_pretrained("google/owlvit-base-patch32")
owlvit_det_model = OwlViTForObjectDetection.from_pretrained("google/owlvit-base-patch32").to(device)
# BirdSoup mean std
mean = [0.48168647, 0.49244233, 0.42851609]
std = [0.18656386, 0.18614962, 0.19659419]
owlvit_det_processor.image_processor.image_mean = mean
owlvit_det_processor.image_processor.image_std = std
# load finetuned owl-vit model
weight_dict = {"loss_ce": 0, "loss_bbox": 0, "loss_giou": 0,
"loss_sym_box_label": 0, "loss_xclip": 0}
model = OwlViTForClassification(owlvit_det_model=owlvit_det_model, num_classes=n_classes, device=device, weight_dict=weight_dict, logits_from_teacher=use_teacher_logits, custom_box_head=custom_box_head)
if model_path is not None:
ckpt = torch.load(model_path, map_location='cpu')
model.load_state_dict(ckpt, strict=False)
model.to(device)
return model, owlvit_det_processor